HOW WASHINGTON CREW GOT GOOD, THRIFTY SEISMIC IN BAD DATA AREA

Craig M. Jarchow
Stanford University
Stanford, Calif.

Rufus D. Catchings, William J. Lutter
U.S. Geological Survey
Menlo Park, Calif.

Countless mature, emerging, and frontier basins remain incompletely tested for hydrocarbons because aspects of their geology make it nearly impossible to acquire clear subsurface seismic images.

One need only look as far as the most productive hydrocarbon provinces in the U.S. for examples of this problem.

In parts of West Texas, most particularly the Val Verde basin, carbonates severely degrade seismic images.

Salt is a well known problem in the Gulf of Mexico.

In Nevada, volcanics make sorting out complicated structure a daunting exercise.

Permafrost is a persistent headache for explorationists working on the North Slope of Alaska.

Although degraded seismic images are commonplace and the lithologies associated with them are numerous, the physical cause of the degradation is roughly the same from basin to basin and lithology to lithology. In nearly all cases, problematic lithologies introduce into a stratigraphic package an abundance of very large seismic velocity contrasts.

In volcanic flows, large velocity contrasts are associated with interbeds, brecciation, and vessiculation. In carbonates, karst features play an analogous role.

When seismic waves impinge upon these contrasts, multiple generation, mode conversion, and scattering occur, producing high-amplitude reverberations (Fig. 1).

Commonly, these reverberations have sufficient strength and duration to completely obscure other seismic reflections that may be of economic importance.

The daunting challenge faced by the exploration geophysicist in bad data areas is to mitigate the effects of problematic lithologies without adversely affecting the cost per kilometer of a seismic program.

The strategy employed in most cases is to "tune" source and receiver arrays to minimize reverberation and hope that any residual reverberation can be removed in processing. Although this strategy is prudent and often succeeds, it doesn't work as often as one would like.

If surface conditions change along a seismic line, source and receiver arrays can quickly go out of tune. Moreover, logistical considerations often place severe constraints on possible array geometries.

If the problematic lithologies are near the surface, reverberations can overwhelm recording equipment, causing deeper reflections of economic interest to fall outside the range of recorded signals.

What can be done to improve this situation? Two approaches are evident to the authors.

The most common approach is to stick with the standard strategy but try harder (read: spend more money). One may use exotic sources, elaborate source and receiver arrays, higher multiplicity, multicomponent receivers, deeply buried receivers, etc.

An alternative approach, to be expounded here, is to simply lengthen the recording spread so that the wide angle portion of the seismic wavefield is recorded. At wide angles, reverberations typically are less of a problem and deeper reflections normally increase in strength due to the well known amplitude vs. offset (AVO) phenomenon.

Although this method probably is not univerally applicable, it was quite successful in the Columbia basin, Washington state, perhaps the most notorious of all bad data areas. This case history is summarized in this article.

COLUMBIA BASIN PROGRAM

The Columbia basin contains thick sequences of primarily Eocene siliciclastic sediments that have significant hydrocarbon potential.

Nearly all wildcat wells drilled into these sediments have had gas shows. One well tested multiple zones with a cumulative flow rate of 5 MMcfd with distillate.

Despite these favorable indications, the Columbia basin remains one of the least explored large onshore basins in the U.S. The reason for this is clear. Covering the prospective sediments of the basin is one of the most voluminous outpourings of continental flood basalt on earth.

These basalts are riddled with large seismic velocity contrasts produced by a thick surface loss, sedimentary interbeds, zones of brecciation, and extensive vessiculation. These contrasts result in highly reverberative seismic data that are typically devoid of interpretable events from the sedimentary section.

In lieu of reliable seismic information, explorationists in the Columbia basin have relied heavily on other geophysical techniques, particularly gravity and magnetotellurics. This has proven to be hazardous. At one wildcat well, these alternative methods mislocated the base of basalt by more than 1 km.

In another well, touted to be the first based solely on geophysical information, crystalline basement was encountered at shallow depth after drilling through a thick basalt section and an embarrassingly thin sedimentary section. Following a series of very expensive ($10 million plus) dry holes such as these two just mentioned, oil companies cut their losses and ceased exploration in the Columbia basin.

Following these setbacks, the authors conducted a careful study of the seismic data acquisition problems in the Columbia basin and concluded that it is possible to acquire reliable seismic information there.

Instead of relying solely on near offset reflections (as is standard practice), the authors suggested that sufficient geologic information for the Columbia basin play exists in other, commonly ignored, portions of the seismic wavefield. One simply harvests information from more fertile portions of the wavefield as opposed to coercing marginal information from near offsets via exotic and expensive shooting and processing methods.

The authors proposed testing these ideas by shooting a seismic profile in the Columbia basin and received support in the endeavor from Amoco Production Co., Chevron Corp., Conoco Inc., Exxon Co. U.S.A., Gas Research Institute, Hunt Oil Co., Meridian Oil Inc., Mobil Oil Corp., Occidental Petroleum Corp., Oryx Energy Co., Shell Oil Co., and Unocal Corp.

The seismic profile was located in central Washington near the cities of Wenatchee, Yakima, and Ellensburg (Fig. 2). Large explosive sources were used to shoot through an 80 km long receiver array.

The entire array was live for each shot, ensuring that wide angle information was recorded. Total cost per kilometer for this line was approximately $3,000 (this number includes some academic discounts).

SEISMIC RESULTS

Despite horrible shooting conditions (a snowstorm with high winds) and massive instrument failures, the data collected along the profile seem to validate this approach (Fig. 3).

At shot-receiver offsets less than about 18 km, reverberations dominate, and no primary reflections can be confidently identified. Beyond 18 km, at offsets not commonly recorded, a reflection from the sediment-basement interface suddenly becomes patently obvious. Existing migration algorithms can be applied to this reflection to arrive at a structural model for the basement.

For reasons beyond the scope of this article, developing an accurate basement model is a crucial aspect of the Columbia basin play.

Other crucial targets in the Columbia basin play are the basalt-sediment interface and structures in the basalt itself. The authors have clearly delineated these features by applying an inversion algorithm to the basalt diving wave (Fig. 3).

Ironically, diving waves are regarded as noise in standard seismic processing, yet they are an important phase for imaging the basalts of the Columbia basin.

Depth to the basalt-sediment interface matches the depths observed in two wells that tie to the seismic profile. In addition, the faults imaged in the basalt correlate with large surface anticlines in the area.

To summarize, the authors used two commonly discarded portions of the seismic wavefield to delineate crucial geologic features of the Columbia basin. Expensive coercion of the near-offset wavefield was avoided.

It is beyond the scope of this short article to delve more deeply into the technical details of the method used, but such information will be discussed fully during a sponsor's workshop scheduled for early July 1991 at Stanford University.

Rather than communicate technical details, this article is designed to make explorationists cognizant of an inexpensive solution to bad-data problems. Perhaps this solution will give some abandoned or marginal plays renewed economic viability.

ACKNOWLEDGMENTS

The authors thank Amoco, Chevron, Conoco, Exxon, GRI, Hunt, Meridian, Mobil, Occidental, Oryx Shell, and Unocal for their support of this program. Amoco in Denver generously supplied extra instrumentation. Digicon's patience and flexibility in the field helped make the program a success.